Trustees bad at picking managers, or managers bad at generating returns?

There's a piece on IPE.com today repeating a point that has been many times over the past few years in the pensions industry. This is that pension funds tend to pick well-performing fund managers at the point at which their performance deteriorates. They also tend to sack poorly-performing managers just when they are about to turn the corner.

One of the key paragraphs in the article, in my view, is this one:

MacDougall explained where managers have performed well in the recent years "it is quite a high likelihood that they actually won’t maintain that performance in the period following when the trustees want to hire that manager.”

Or put it another way - manager performance is mean reverting.

Further down the article is this line -

His sentiments were echoed in a recent 'What if?' market review by consultants Watson Wyatt which warned managers should not be chosen judged on past performance but only on skills.

Pretty much an indication that manager selection is a leap of faith, rather than something you can do using stats.

I wonder how many times we have to hear versions of these basic truths before we conclude that active management is basically a waste of money. No managers seem able to deliver prolonged outperformance, and statistically those that achieve it for some time might just be reflecting the random nature of market. Maybe we need active management only to play the arbitrage function in keeping prices reasonably rational (although of course this didn't happen in the TMT bubble).

In which case can't we have it much cheaper please?